The business case for a robot that moves cells between machines

A specialised instrument can process living cells and still leave a person to connect the bag, move the liquid and prepare the next step. Multiply Labs builds robots to connect those operations. For therapies made from a patient’s own cells, every patient requires a separate batch.[2]

The company announced a $75 million Series B on 6 October, led by Patrick Soon-Shiong with NantWorks and joined by investors including AstraZeneca. It says the money will expand capacity and its engineering, regulatory and commercial teams as it moves from clinical deployments towards commercial production. The product is a cluster of robots and instruments owned and operated by the pharmaceutical customer. It promotes a 74% reduction in cost per dose.[1]

Automating the spaces between instruments

A 2026 prototype puts a robot arm on a central rail between laboratory modules. External adapters grip disposable parts without entering the sterile fluid path. Force sensors check connections, scales track transfers and software records each batch.[2]

That arrangement has an economic attraction. A room or instrument costs money while it waits as well as while it works. Adding people can increase the number of handoffs handled at once, but adds staffing costs. Scheduling a shared collection of machines offers a different way to increase useful work: reduce the time each part spends waiting for something else.

Researchers captured a scientist’s movement while mixing settled cells back into liquid and translated it into robot motion. Broader imitation and reinforcement learning remained future work. It reproduced a skilled movement rather than independently learning a manufacturing process.[2]

The cost claim has an instructive breakdown

A 2024 presentation by Multiply co-founder and researcher Alice Melocchi shows an earlier economic model with the same rounded reduction. It compares a traditional manual suite with a robotic baseline. These are scenario costs, not savings measured across a commercial factory.[3]

Labour, facility and hardware costs shrink dramatically in the model, while consumables become more expensive. Calculated from the slide, consumables account for about 92% of the robotic total. Under those assumptions, removing handling and spreading fixed costs across more production shifts the dominant expense towards materials used for each batch.[3]

That distinction explains what the robots are being paid to achieve. A bag or reagent remains a cost each time a treatment is made. A room, instrument or robotic system can serve more batches if its productive use rises. The model makes efficient use of those shared resources central to the return. Its dollar totals cannot be carried unchanged into a different process, staffing plan or production schedule.

The company’s other large claim also needs its unit. In a 2025 Universal Robots account, the asserted improvement of up to 100 times refers to patient doses per square foot of cleanroom space. Multiply’s supplier describes stacking equipment vertically and using robots to work across it. The measure is about production density; it does not establish that cells grow 100 times faster or that an individual treatment takes one-hundredth as long.[4]

What the newer experiments establish

Across three bag-transfer experiments, an average 96% of viable cells reached the destination bag. Several tasks produced results similar to human operators’ results, but robotic handling produced lower average counts in one low-concentration counting test.[2]

People preassembled and inspected consumables. Final formulation, filling and cryopreservation remained to be automated, and the study did not characterise a complete therapy’s function.[2]

Multiply is selling pharmaceutical companies a way to bring that coordination into their own production. The customer retains the manufacturing operation; Multiply supplies the machinery that connects its steps. The investment case turns on whether a repeatable set of transfers, measurements and scheduled actions can let the same specialised facilities serve more patients. The useful output is a completed batch moving through a dependable process.[1]

Report an error or suggest a correction